Implement Token Truncation for GPT-4/GPT-3.5 Context Overflow

[RoyLeviGit]

In instances where the GPT-4/GPT-3.5 context exceeds its limit, we need to implement a feature to automatically shorten tokens. This should utilize the session_memories concept.